I don't think any C compiler for 51' will support 128 bit Most 51' C compiler support int (16 bit) some better one support float and long int 32 bit) with library for double(64 bit).

You have to write your own routine for 128 bit arth in assem. You can pick up any 16 or 32 bit add , sub , mult & divi routine from numerous 51' assembly math routine libray and easily change them for 64 bit or 128 bit manipl.
But I want to ask same question as Mr. Kleiber for which appl you want 128 bit precision??